agentleFS
arnaudgelasarnaudgelas/K-Ops/.claude/skills/research-workflow/SKILL.md

research-workflow

Skill1 starsChanged 7 months agoMITView on GitHub

Manage resumable research runs under `research/` and coordinate the collect, review, report, import, and archive steps.

14 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 8 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run higher-stakes topics as resumable, tiered research threads outside the curated vault until they are ready to promote.

    line 9
  2. Very specific:

    Keep active run files in research/, not in notes/.

    line 33
  3. Distinctive:

    Treat imported model-generated reports as leads, not authority.

    line 34

“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Research Workflow

Goal

Run higher-stakes topics as resumable, tiered research threads outside the curated vault until they are ready to promote.

Use When

  • The user wants a structured research run rather than an immediate vault answer.
  • A topic needs source collection, contrarian review, and a final report.
  • An external model report should be imported as a lead, not treated as authority.

Core Flow

  1. Start a run: uv run kops research-start --topic "<text>" --tier <fast|standard|deep>.
  2. Check status: uv run kops research-status [topic|all].
  3. Collect sources: uv run kops research-collect --agent <codex|claude|gemini> --topic "<topic>" (runs collect prompt).
  4. Adversarial review: uv run kops research-review --agent <codex|claude|gemini> --topic "<topic>" (runs review prompt).
  5. Draft report: uv run kops research-report --agent <codex|claude|gemini> --topic "<topic>" (runs report prompt).
  6. Import external reports: uv run kops research-import --topic "<topic>" --path <file> --provider <provider> --origin <origin>.
  7. Archive completed run: uv run kops research-archive --topic "<topic>" (after phase shows done).

Phase Guidance

The research-collect, research-review, and research-report CLI subcommands render their own runtime prompts (from kops/templates/), so no separate skill is needed. When running or reviewing a phase by hand, hold to these contracts:

  • Collect (research-collect): populate the findings file with high-signal, sourced claims. Create/update source notes in notes/Sources/ under the strict metadata schemas; prefer primary sources; mark imported model reports authority: lead_only + verification_state: needs_primary_sources. Distinguish evidence from inference; keep uncertainty explicit; append a progress log.
  • Review (research-review): adversarial contrarian pass — try to break the thesis, not confirm it. Populate ## Strongest Objections, ## Missing Evidence, ## Claims To Soften with genuine gaps (never fabricated counter-evidence); name missing primary sources to consult before drafting.
  • Report (research-report): synthesize brief + sources + findings + contrarian review into a bounded report. Build from collected evidence only, address the strongest objections, keep conclusions bounded by evidence, cite source summaries with wikilinks, and distinguish established from provisional findings.

Rules

  • Keep active run files in research/, not in notes/.
  • Treat imported model-generated reports as leads, not authority.
  • Preserve the status/progress/review/report linkage across steps.
  • Promote durable findings into the main vault only after verification.

Outputs

  • research/briefs/ (brief files: type: research-brief)
  • research/notes/ (status: type: research-status, progress: type: research-progress, review: type: research-review)
  • research/findings/ (findings: type: research-findings)
  • research/reports/ (reports: type: research-report)
  • research/imports/ (manifest: type: research-import-manifest)
  • research/archive/ (manifest: type: research-archive-manifest)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.